The effect of stress andsatisfaction on productivity
George Halkos and Dimitrios BousinakisDepartment of Economics, University of Thessaly, Volos, Greece
Abstract
Purpose – This study aims to investigate the effects of stress and job satisfaction on the functioningof a company. It seeks to focus on factors affecting stress and job satisfaction such as number of workhours, good relations between management and employees, good function of the group and workrelated to employees’ area of education.
Design/methodology/approach – A random sample of 425 employees in the private and publicsector and two stage cluster sampling is first used to collect primary data. Factor analysis is used nextto identify the responsible factors for the correlation among a large number of qualitative andquantitative variables and their influence on productivity. Logistic regression is used next presentingmany useful elements concerning the function of stress, satisfaction and supportive elements onproductivity.
Findings – As expected, increased stress leads to reduced productivity and increased satisfactionleads to increased productivity. When work begins to overlap with workers’ personal life this implies anegative effect on productivity. Quality work is more related to conscientiousness and personalsatisfaction than work load. Energetic and active individuals affect productivity positively.
Originality/value – The paper presents a number of qualitative variables as factors representingstress and satisfaction. This is achieved using factor analysis. Next logistic regression offers the oddsratio and the corresponding probability of the effect on productivity after a change in stress andsatisfaction. The empirical analysis completes the existing literature contrasting different theoreticalsets of predictor variables and examining their effect on productivity. Additionally, in the study thestates of stress and job satisfaction are the result of the interaction of the environment’s demands withthe personal characteristics.
Keywords Stress, Job satisfaction, Productivity rate, Factor analysis, Greece
Paper type Research paper
IntroductionTwo important problems that modern organizations are faced with are stress and jobsatisfaction of their employees. At a first look we could deduce that these two problemsare not correlated. But if we look at these issues in depth we see that one affects theother and if both function well it could lead to positive results for employees’ work andorganization.
Stress can be considered as an unpleasant emotional situation that we experiencewhen requirements (work-related or not) cannot be counter-balanced with our ability toresolve them. This results in emotional changes as a reaction to this danger. It stems
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1741-0401.htm
The authors thank the ICAP Group for providing the database on the companies surveyed in thework. Thanks are due to an anonymous referee, to the editorial team and to John Heap for thevery helpful, detailed and constructive comments raised on an earlier draft of the paper. Theauthors also thank Richard Dawson for improving the overall fluency of the text. Any remainingerrors are the authors’ responsibility.
The effect ofstress and
satisfaction
415
Received May 2008Revised July 2009
October 2009Accepted November 2009
International Journal of Productivityand Performance Management
Vol. 59 No. 5, 2010pp. 415-431
q Emerald Group Publishing Limited1741-0401
DOI 10.1108/17410401011052869
from the relationship between a person and its environment and it appears as pressurethat is subjective because the same stressors can affect one person but not another.When an employee can manage the pressures of the job and the possibility to completea task is substantial then stress can work as a motivating factor.
Satisfaction is a regulating factor for stress. Theories during the neo-classical period(1920-1950) supported the proposition that employee satisfaction directly affectsproductivity. They believed that there existed a cause-effect relationship betweensatisfaction and productivity. This was the reason why organizations used variousmeans in order to increase employee productivity and thus increase productivity.
There is no doubt that in many cases productivity has to do with factors which areexternal to the person, but affect performance (e.g. the performance of a salesperson isclosely linked with market mobility, despite the persuasion dynamics he/she maypossess). In many cases, also, work performance of an individual is directly linked tothe performance of other employees in the same space, so the individual cannot sethis/her own standards, especially if there are some informal social rules.
It is believed that job satisfaction is directly correlated with the mental health of theworkforce and the organizations’ interest in high productivity and a stable andpermanent workforce. Stress on the other hand is the main cause of problems not onlyin persons’ professional but also in their personal lives. It can also create physical andpsychosomatic symptoms. A stress-filled employee makes wrong decisions and hasnegative relationships with co-workers. Both these elements can bear a negativeoutcome in the productivity of a group thus creating an added cost to a company.Reduced productivity, mistakes, low quality work and absenteeism are signs of astressed employee.
On the other hand a satisfied employee is a vital prerequisite for a healthy company.Work related stress is a vital factor to job satisfaction. When it functions as a motivatorthen it results in creativity and satisfaction and consequently dissolves boredom andmundaneness. When stress functions as a negative factor it results to aggression andin low job satisfaction. Job satisfaction can lead to prevention of stressors though jobincentives.
In this study our effort focuses on the investigation and analysis of the effect of thequality factors of stress and satisfaction on productivity. Using two-stage clustersampling and a random sample of 425 employees in the private and public sector weextract two factors representing stress and job satisfaction and we investigate theireffects on the functioning of a company and/or organizations. We focus our attentionon creativity, group activity and independent work, factors that affect stress and jobsatisfaction. Here, the state of stress is a result of the interaction of the environment’sdemands with the personal characteristics. Specifically, factors affecting creativity andproductivity are the number of work hours, good relations between management andemployees, good function of the group and work related to employees’ area ofeducation. Independent work increases job satisfaction and productivity of a person. Itworks as reducing stress.
The structure of the paper is the following. The next section reviews the existingrelative literature. The next two sections present the sampling framework and theadopted methodology for the analysis of the data collected. Then the empirical resultsderived are presented and discussed. The last section concludes the paper andcomments on the policy implications of our empirical findings.
IJPPM59,5
416
Existing studies in the literatureMany attempts have been made to interpret and define stress. The first theory onstress belongs to Freud (1978), who considered stress as the result of reduced dischargeof libidinal energy, either due to external obstacles or due to internal ones. In the 1960s,the cognitive approach to the personality was created, which considers that stress iscreated when the individual is not capable or believes that he/she is not capable ofmeeting the demands of a certain situation, and that these situations are a threat to theindividual’s health.
Aldwin (1994) considers that stress refers to the experience created as a result of theinteraction of the individual and the work environment. This interaction may lead topsychological and physiological tension. Selye (1964) defines stress as the naturaldegeneration of the body and as the non-specific response of the body to any demandplaced on it. He himself recognised the meaning of positive stress, which not only doesnot cause degeneration and malfunctions, but can also act as a productive factor and asa factor of development and creation.
Karasek (1979) proposed a theoretical model, where the basic factors that causestress to the employee are:
(1) The work or project the employee is called to put into effect in itself.
(2) The limits of initiative taken by the employee, the independence and themargins of control he/she has in the job.
(3) Social relations with seniors, colleagues and subordinates.
The existence of just one of these three factors is not enough to create stress. All threetogether, however, definitely affect the employee.
Warr (1990) considers that each of Karasek’s work factors must exist at anappropriate analogy so as not to create stress. As stressful as not having muchinitiative margin may be, extremely large margins are equally stressful. According toWarr, the basic factors burdening stress are decision-making and the development ofknowledge, abilities and experiences, satisfactory remuneration, working duties thatare interesting and varied, precise roles, physical safety, tangible targets, socialrecognition and the potential for interpersonal communication.
According to Siegrist (1996), there must be a balance between what employees“invest” in the job and what they get back. In opposite cases, they feel oppressed anddissatisfied. The term effort contains two dimensions, exogenous and endogenous. Theformer concerns the effort that the employees make in order to fulfil their workingduties, while the latter concerns the internal motives that urge them to perform (e.g. theneed for social recognition, etc.). As reciprocation, employees get financialremuneration from the job, the potential to sustain or upgrade their workingposition, expectation satisfaction, security etc.
There are many theories that have dealt with satisfaction; some of them are in thesame lines while others differ greatly from each other. Initially, Maslow (1954)supported the anthropocentric function of organisations, with the existence of ahierarchy of various need forms. Initially we have physiological needs, which areclearly biological, such as food, clothing, accommodation etc. These constitute the basefor the individual to move on to the satisfaction of psychological needs. When physicalneeds are satisfied, then the needs of safety or certainty arise. These include the needfor stability, protection from dangers, and provision for the future.
The effect ofstress and
satisfaction
417
A number of researchers have found a connection between intention to leave one’sjob and job dissatisfaction (Heslop et al., 2002; Brief and Weiss, 2002; Clugston, 2000).Halpern (1999) claims that employee turnover caused by job dissatisfaction has causedcompany costs in terms of recruitment, selection and training new employees.Researchers have also studied job satisfaction in a wide range of professions likeindustrial teacher educators (Brewer and McMahan-Landers, 2003, a, b), teachers(Bogler, 2002), physicians (Bergus et al., 2001), customer service employees (Carless,2004), student support personnel (Brewer and Clippard, 2002), youth developmentorganizations (Petty et al., 2005) and management of healthcare workforce (Labiriset al., 2008).
When safety needs are satisfied, then social needs arise. As a social being, theindividual needs to belong in groups, to have loving relationships with otherindividuals, friendships, etc. Estimation needs constitute a development of socialneeds, because here the individual does not only desire to belong to a group, but also tobe recognised, appreciated and respected by others. The satisfaction of these needscreates self-confidence, power and prestige. Finally, there is the need for self-realisationthat is the need for maximising potential towards higher forms of action. The desire isto become what somebody can become, and this state, of completion, is reached by fewpeople.
Alderfer (1972) amended Maslow’s theory and supported the contention that if forsome reason the individual cannot satisfy his/her needs on a higher level, then he/shereturns to the needs of a lower level that are already satisfied. Through his ERG theory(existence, relatedness, and growth), Alberfer sorted the needs into three categories:
(1) Existence (here we find Maslow’s physical and safety needs).
(2) Relatedness (Maslow’s social needs).
(3) Growth (Marlow’s estimation and self-realisation needs).
Finally we must mention Herzberg’s (1966) theory in passing as we believe that itconstituted the base for the development of several theories. In Herzberg’s theory wefind two different kinds of factors, motivators and hygiene factors, which are related towork satisfaction. According to Herzberg, positive stances towards work which lead tosatisfaction are related to the work content, e.g. achievement, recognition,responsibility, development potential, and the nature of the work. These factorswere named motivators as they contribute to the urging of the individual towardsgreater performance and effort.
On the other hand, negative stances that lead to dissatisfaction are connected to theframework of the organisation, such as management, supervision, remuneration,interpersonal relationships. These factors were named hygiene factors, as theycontribute to the prevention of work dissatisfaction, while their effect on the creation ofpositive feelings is very limited.
With reference to the relationship of satisfaction with productivity and based on theassumption that there is a relationship, Porter and Lawler (1986) created a model inorder to examine the matter of activation. The model is based on the assumption thatrewards create satisfaction and that some times performance leads to remuneration ofvarious kinds, which create satisfaction in workers. Thus, productivity is related tosatisfaction through the notion of rewards and therefore comes into contrast with theneo-classical approach, which considered satisfaction a cause and prerequisite for good
IJPPM59,5
418
performance. There are many factors that lead to the view that the satisfied worker isnot necessarily a productive one.
Locke (1976) considers that the relationship between satisfaction and productivity isreciprocal. It is not, thus, satisfaction that leads to productivity, but productivity thatleads to satisfaction. Then, satisfaction affects productivity mainly in an indirect way,creating a feeling of dedication towards the organisation and its targets. Beyond thisrelationship of productivity-satisfaction-productivity, it is possible to have a secondaryincrease of satisfaction, provided that productivity results in the increase of otherremunerations related to work (promotion, authority, bonus, etc) that contribute to theincrease of satisfaction.
Finally, with the development of new technologies and the globalization of economicgrowth a number of changes in the labour market have been experienced with eitherrelatively advantaged and stable employment or uncertain employment characterizedby volatility and low salaries (Ferrie et al., 1999; Paoli, 1997). Structural unemployment,underemployment and early retirements have increased and continue to increaseleading to increased stress, job insecurity and lower job satisfaction.
Approach to measurement development: the questionnaireIn our study, apart from stress and satisfaction levels that interest us, we made aneffort to collect information concerning the parameters related to these elements, eitherseparately or as a whole. A number of variables were considered such as thesocio-economic (age, marital status, income, sex) and other qualitative (that givenon-numerical information) variables like “creativity within the organization”, “labouraccountability”, “higher rewards” and the level of education.
Relying on the existing literature a new questionnaire was developed and was firsttried on 20 employees (around 5 per cent of the final sample). A number ofmodifications were made before the final version. Testing the reliability of ourinstrument a Cronbach alpha coefficient of 0.92 was estimated. This coefficient showshow all the statements of the questionnaire relate to one another in content.
The data collection was performed within one month and solely by means ofpersonal interviews. Participants replied to a number of statements using a five-pointLikert scale with 1 corresponding to “very little” and 5 to “very much”. In case ofnegative statements we had to reverse the scores with the value of 1 corresponding to“very much” and the value of 5 corresponding to “very little”. The ones that answeredat the extreme ends (“very little” or “little”) seem not to consider important the effect ofchanges on productivity, whereas the ones that answered “not much”, “very” and “verymuch” seem to consider the effect as important. The ones that did not reply wereexcluded from the analysis of the sample.
The adopted methodologyIn order to analyse the relationship between stress, satisfaction and productivity, weperformed a study on a random sample of 425 individuals. The population of our studyconsists of employees working in private enterprises and public organisations fromthroughout the country (excluding non-profit organizations). A list of all companiesoperating in Greece was provided by ICAP and this list was our sampling frame. Asummary of this information is presented in Table I. The entire population was used in
The effect ofstress and
satisfaction
419
order for the sample to be representative, random and as large as possible (Gay andAirasian, 2003).
Specifically, our sample initially contained primary units and then through themsecondary units was selected. Our work was based on the method of two-stage clustersampling and not on a single stage sampling, such as random, systematic or stratified.Cluster sampling requires the division of the population into groups ofelements/clusters in such a way that each element belongs to one and only onecluster. We preferred cluster sampling instead of stratified as the former tends toprovide better results when the elements within the cluster are heterogeneous. We haveadopted a two stage cluster sampling and developed first a frame consisting of allemployees in private and public sectors in middle and high positions. We have selectedfirst with the use of random numbers a random sample of 94 companies and then arandom sample from each of the 94 sampled clusters.
Factor analysis is used first to group the variables (see Table II) into main factorsaccording to their impact similarity and avoiding the problem of multicollinearity. Theidea of performing a factor analysis came from the fact that some variables areexpected to present an increased correlation as a result of overlapping variationbetween them. That would result in multicollinearity in a multiple regression modelsetup. Researchers suggest the application of factor analysis in order to examine thestructure of the overlapping variation between the predictors (Leeflang et al., 2000)claiming that the only problem in this case remains the theoretical interpretation of thefinal components (Greene, 2000; Gurmu et al., 1999).
Sector CodeTotal numberof companies
Total numberof employees
Sampledcompanies
Sampledemployees
Agriculture and related A-B 01 249 3,300 3 14Fish farming A-B 05 114 3,920 2 10Manufacturing (food-drinks) D 15 1,214 62,626 5 16Manufacturing (Tobacco) D 16 4 1,973 1 2Manufacturing (textile-leather) D 19 73 2,010 1 2Manufacturing (chemicals) D 24 286 22,362 3 14Constructions F 45 1,879 36,066 3 12Retail trade G 52 1,304 101,561 7 24Cars trader – leasing G 50 906 23,864 5 21Telecommunication – internet I 64 115 35,867 4 19Financial services J 65 25 74,127 9 25Insurance J 66 239 8,829 3 14Vehicles and equipment KMNO 71 266 3,027 4 17Information technology KMNO 72 464 11,932 2 11Other sectors KMNO 74 1,967 65,061 9 39Education KMNO 80 244 9,066 2 8Health KMNO 85 342 20,453 7 37Total private companies 9,691 486,044 70 285Total public companies 24 140Total 94 425
Source: ICAP (personal communication)
Table I.The sampling frame usedin our analysis
IJPPM59,5
420
Com
pon
ent
mat
rix
Rot
ated
com
pon
ent
mat
rix
Var
iab
les
Fac
tor
1F
acto
r2
Fac
tor
1F
acto
r2
Com
mu
nal
itie
s
Wor
kca
ncr
eate
dif
ficu
ltie
sin
per
son
alli
fe0.
381
3.68
6E-0
20.
350
20.
155
0.14
6W
ork
stre
ssor
sca
naf
fect
the
rest
0.49
70.
118
0.49
12
0.14
10.
261
Wor
klo
adu
pse
tsp
eop
le0.
450
0.29
30.
536
3.46
9E-0
20.
289
Str
ess
from
tran
sfer
toor
from
wor
k0.
395
0.17
60.
431
24.
02E
-02
0.18
7S
tres
sfr
omth
ew
ork
env
iron
men
t0.
390
0.10
50.
391
29.
96E
-02
0.16
3S
tres
sab
out
wor
kh
ours
0.50
80.
288
0.58
41.
620E
-03
0.34
1S
tres
sab
out
job
secu
rity
0.44
10.
218
0.49
12
2.61
E-0
20.
242
Str
ess
from
coop
erat
ion
-com
mu
nic
atio
nw
ith
oth
ers
0.59
30.
280
0.65
42
4.67
E-0
20.
430
Str
ess
from
lack
ofcr
eati
vit
y0.
689
0.27
90.
738
29.
47E
-02
0.55
3S
tres
sab
out
chan
cefu
rth
ered
uca
tion
0.52
80.
246
0.58
12
4.47
E-0
20.
339
Str
ess
abou
tp
erso
nal
wor
km
eth
od0.
547
0.25
70.
602
24.
44E
-02
0.36
5S
tres
ssu
rrou
nd
ing
wor
kre
lati
ons
0.54
50.
376
0.66
06.
075E
-02
0.43
9S
tres
sab
out
sala
ry0.
569
0.17
50.
582
20.
126
0.35
4S
tres
sab
out
uti
liza
tion
ofk
now
led
ge
–ca
pab
ilit
ies
0.62
90.
242
0.66
72
9.78
E-0
20.
455
Str
ess
abou
tre
lati
ons
wit
hm
anag
emen
t-le
ader
ship
0.70
40.
236
0.72
92
0.14
00.
552
Str
ess
abou
tch
ance
sof
evol
uti
on0.
590
0.23
90.
631
28.
07E
-02
0.40
5S
tres
sab
out
man
agem
ent
pol
icy
0.70
00.
212
0.71
42
0.15
80.
534
Str
ess
abou
tch
ang
em
anag
emen
t0.
532
0.25
10.
587
24.
26E
-02
0.34
6S
atis
fact
ion
from
the
org
aniz
atio
n0.
429
20.
343
0.20
52
0.50
90.
302
Sat
isfa
ctio
nfr
omw
ork
role
0.36
22
0.42
00.
109
20.
544
0.30
8S
atis
fact
ion
from
the
wor
ken
vir
onm
ent
20.
266
0.31
12
7.96
E-0
20.
402
0.16
8S
atis
fact
ion
from
coop
erat
ion
-com
mu
nic
atio
nw
ith
oth
ers
20.
439
0.27
82
0.24
60.
457
0.27
0S
atis
fact
ion
from
lack
ofcr
eati
vit
y2
0.35
30.
633
2.66
5E-0
30.
725
0.52
6S
atis
fact
ion
abou
tch
ance
furt
her
edu
cati
on2
0.30
50.
489
22.
58E
-02
0.57
60.
332
Sat
isfa
ctio
nab
out
per
son
alw
ork
met
hod
20.
344
0.50
82
5.11
E-0
20.
611
0.37
6S
atis
fact
ion
surr
oun
din
gw
ork
rela
tion
s2
0.21
60.
116
20.
131
0.20
75.
999E
-02
Sat
isfa
ctio
nab
out
sala
ry2
0.43
70.
446
20.
162
0.60
30.
389
Sat
isfa
ctio
nab
out
ofu
tili
zati
onk
now
led
ge
–ca
pab
ilit
ies
20.
445
0.69
92
4.53
E-0
20.
827
0.68
7S
atis
fact
ion
abou
tre
lati
ons
wit
hm
anag
emen
t-le
ader
ship
20.
417
0.48
52
0.12
50.
627
0.40
9S
atis
fact
ion
abou
tch
ance
sof
evol
uti
on2
0.36
60.
586
23.
13E
-02
0.69
00.
478
Sat
isfa
ctio
nab
out
man
agem
ent
pol
icy
20.
384
0.53
52
7.21
E-0
20.
654
0.43
3S
atis
fact
ion
abou
tch
ang
em
anag
emen
t2
0.10
00.
262
4.09
8E-0
20.
277
7.84
6E-0
2C
um
ula
tiv
ep
rop
orti
onof
tota
lsa
mp
lev
aria
nce
exp
lain
ed65
.663
.363
.3K
MO
0.86
0B
artl
ett’
ste
stof
sph
eric
ity
5.56
0.3
(Sig
.¼0.
000)
Table II.Factor analysis results
The effect ofstress and
satisfaction
421
Specifically referring to the factor model, the factor scores are calculated as:
F̂ ¼ XB̂ ð1Þ
where F̂ is an mxn matrix of m factor scores for n indicators, X is an nxp matrix ofobserved variables and B̂ is a pxm matrix of estimated factor score coefficients. In thePrincipal Components method applied here for the extraction of the factors the scoresare exactly calculated. Residuals are computed between observed and reproducedcorrelations.
If the common factors F and the specific factors u can be assumed normallydistributed, then maximum likelihood estimates of the factor loadings and specificvariances may be obtained. In our case we have followed the varimax rotation. Theobjective of this rotation is to determine the transformation matrix in such a way asany given factor will have some variables loaded high on it and some loaded low on it.This may be achieved by maximizing the variance of the square loading acrossvariables subject to the constraint that the communalities of each variable remain thesame (Johnson and Wichern, 1998; Sharma, 1996).
Logistic regressionThe idea of performing a regression analysis between a dependent variable andextracted factors is not a new one. Dunteman (1989) also suggests this process to copewith multicollinearity in a regression analysis model and it is also an indicated way tominimize the number of independent variables and maximize the degrees of freedom.
After presenting the basic variables and the corresponding answers of theinterviewees, we will proceed with the sampling of the effect of changes to productivitybased on those variables. More specifically as a dependent variable will use the effectof stress and satisfaction to productivity. As independent variables were consideredthe socio-economic as well as various other qualitative variables mentioned above.Various dummy variables were constructed in relation to the ranking within theorganization (employee, supervisor, manager) as well as the impact on productivitybased on different age groups. Those variables were used in a logistic regression.
The method was preferred from the multiple regression as the dependent variable isdichotomous and discontinued. Additionally the logistic regression is the moreappropriate monotonic function for the sample of gathered data compared to thecriterion of least squares of a multiple regression. Also the logistic regression waspreferred from a discriminant analysis since the latter is based on the hypothesis of themultivariate normality and the equal variance-covariance matrices across teams.Those hypotheses are not required in the logistic regression.
Here the dependent variable is a dichotomous variable Y which takes the value 1with probability Q and the value 0 with probability 1-Qi. This means that thelikelihood function is[1].
LðY;QÞ ¼Yn1
i¼1
Qi
! Yni¼n1þ1
ð1 2QiÞ
" #ð2Þ
The logit form of the model is a transformation of the probability Pr(Y ¼ 1) that isdefined as the natural log odds of the event E(Y ¼ 1). That is:
IJPPM59,5
422
logit PrðY ¼ 1 Xj Þ ¼ b0 þXKk¼1
bkXk
The regression coefficients b’s of the proposed logistic model quantifies therelationship of the independent variables to the dependent variable involving theparameter called the odds ratio (OR). As odds we define the ratio of the probability thatimplementation will take place divided by the probability that implementation will nottake place. That is:
Odds ðE Xj 1;X2; . . . ;XnÞ ¼PrðEÞ
1 2 PrðEÞð3Þ
As our main interest is in terms of the main effects we have ignored interactions.Working with the two factors extracted the logit form of the fitted model may berepresented as:
logit ½PrðY ¼ 1Þ� ¼ b0 þ b1 Factor 1 þ b2 Factor 2 þ 11t
where Y denotes the dependent variable as 1 for significant influence of stress andsatisfaction on productivity and 0 for insignificant effect.
Apart from the model formulation using the extracted factors we propose threeother formulations modeling productivity and socioeconomic variables, stress andsatisfaction. Specifically, the first formulation concerns a number of socioeconomicvariables like:
Logit ½PrðY ¼ 1Þ� ¼ g0 þ g1 Age þ g2 Education level þ g3 Work experience
þg4 Distance þ g5 Sector þ g6 Position þ 12t
The other two formulations refer to modelling productivity and stress and satisfactionrespectively. That is:
Logit ½PrðY¼ 1Þ� ¼ d0 þ d1 Hurried þ d2 Low quality þ d3 Effects in private life þ 13t
Logit ½PrðY ¼ 1Þ� ¼ z0 þ z1 Job satisfaction via education þ z2 Job satisfaction via
rightness þ z3 Job satisfaction via qualification þ z4 Job satisfaction via
organization þ 14t
Where eit the disturbance terms and bi, gi, di, zi the parameter estimates[2].
Empirical resultsGoing on to our statistical analysis, Table II presents factor loadings and specificvariance contributions according to maximum likelihood method of extraction in afactor analysis setup. Looking at Table II it can be seen that the first 18 questionsdefine factor 1 (high loadings on factor 1, small or negligible loadings on factor 2) andrepresent stress. The questions refer among others to stress from work to personal lifeand stress from a number of cases like work environment, lack of creativity,
The effect ofstress and
satisfaction
423
surrounding work relations, chances of evolution, change management, managementpolicy etc. Similarly the other 14 questions define factor 2 (high loadings on factor 2,small or negligible loadings on factor 1), which represents satisfaction. The questionsrefer among others to satisfaction from work role, work environment, personal workmethod, surrounding work relations, utilization of knowledge-capabilities, salary etc.
The communalities being high indicate that the two factors account for a largepercentage of the sample variance of each variable and is evidence that the modelpresents stability. From the same table the KMO index is close to unit (0,860) whichimplies that the sum of squares of the partial correlation coefficients between all thepairs of variables is low. This result shows that in our case factor analysis is strong.Similarly, the value of the Bartlett’s test of sphericity is very large (5.560,3) and thelevel of statistical significance is 0.000 leading to the rejection of the null hypothesisthat the matrix of correlation coefficients is unity.
The results of the fitted logistic models are presented in Table III. The individualstatistical significance of the b estimates is presented by the Wald (Chi-square). Thesignificance levels of the individual statistical tests (i.e. the P-values) are presented inparentheses and correspond to Pr . Chi-square.
In the model formulation using the extracted factors as explanatory variables wehave statistical significance for both factors. In the case of using the socioeconomicvariables as independent we see that the variable distance is statistically significant inall levels of significance. Similarly, the variables work experience and educational levelare statistically significant for the levels of 0.05 and 0.1 and the variables workexperience and sector for 0.1. The variable position is statistically insignificant. In thecase of the model with the proposed variables representing stress we see that thevariables low quality and hurried are statistically significant for the levels of 0.05 and0.1 and the variable effect in private life for 0.1. Finally, in the case of the model withthe proposed variables representing satisfaction we see that the variables jobsatisfaction via rightness and qualifications are statistically significant in all levels ofsignificance while the variable job satisfaction via education is significant for the levelof 0.1. The variable job satisfaction via organization is statistically insignificant.
Being more specific, in case we run the model with the socioeconomic variables thenthe coefficient of age is b_1 ¼ 20.636, which implies that the relative risk of thisparticular variable is eb_1 ¼ 0.529 and the corresponding percentage change ise b̂1 2 1 ¼ 20.471. This means that in relation to age the odds of persons’ ability toincrease productivity decreases by almost 47 per cent ceteris paribus. In the case of workexperience b_2 ¼ 20.229, which implies that the relative risk of this particular variableis eb_2 ¼ 1.349 and the corresponding percentage change is e b̂2 2 1 ¼ 0.349. This meansthat in relation to work experience the odds of persons’ ability to increase productivityincreases by almost 0.35 per cent all other remaining fixed. Similarly, the odds of persons’ability to increase productivity decreases by 0.55, 0.29, 0.51 and 0.113 in relation todistance from work, education level, sector of employment and position respectively.
We may compute the difference e b̂i 2 1 which estimates the percentage change(increase or decrease) in the odds p ¼ PrðY ¼ 1Þ=PrðY ¼ 0Þ for every 1 unit in Xi
holding all the other X’s fixed. In case we run the model with the productivity againstthe stress statistically significant variables then the coefficient of someone hurried isb_1 ¼ 0.387, which implies that the relative risk of this particular variable iseb_1 ¼ 1.473 and the corresponding percentage change is e b̂1 2 1 ¼ 0.473. This
IJPPM59,5
424
Variables EstimatesOddsratio Estimates
Oddsratio Estimates
Oddsratio Estimates
Oddsratio
Constant 1.521 7.231 0.537 22.211Wald [100.8] [22.476] [0.301] [36.973]P-value (0.000) (0.000) (0.583) (0.000)
Factor 1 (stress) 20.392 0.676Wald [7.596]P-value (0.006)
Factor 2 (satisfaction) 0.442 1.556Wald [10.616]P-value (0.001)
Age -0.636 0.529Wald [5.529]P-value (0.019)
Work experience 20.299 1.349Wald [3.128]P-value (0.077)
Distance 20.803 0.448Wald [24.511]P-value (0.000)
Education level 20.349 0.705Wald [4.057]P-value (0.044)
Sector 20.720 0.487Wald [2.998]P-value (0.083)
Position 20.120 0.887Wald [1.680]P-value (0.195)
Hurried 0.387 1.473Wald [4.999]P-value (0.025)
Low quality 0.483 1.621Wald [4.558]P-value (0.033)
Effect in private life 20.278 0.757Wald [2.815]P-value (0.093)
Job-satisfaction viaeducation 0.617 1.853
Wald [2.751]P-value (0.097)
Job-satisfaction viarightness 1.130 3.095
Wald [13.45]P-value (0.000)
Job-satisfaction viaqualifications 1.557 4.745
Wald [24.13]P-value (0.000)
Job-satisfaction viaorganization 20.333 0.717
Wald [0.924]P-value (0.336)
Nagelkerke R2 0.1 0.16 0.1 0.34Hosmer Lemeshow 4.478 5.731 9.850 0.922
(0.812) (0.677) (0.276) (0.969)Likelihood ratio 331.357 225.26 246.79 393.904
Table III.The logistic regression
results
The effect ofstress and
satisfaction
425
means that in relation to stress expressed by hurried the odds of persons’ ability toincrease productivity increase by almost 47 per cent ceteris paribus. In the case of lowquality in the work produced b_2 ¼ 0.483, which implies that the relative risk of thisparticular variable is eb_2 ¼ 1.621 and the corresponding percentage change ise b̂2 2 1 ¼ 0.621. This means that in relation to low quality in production the odds ofpersons’ ability to increase productivity increases by almost 0.62 per cent all otherremaining fixed. Finally, the odds of persons’ ability to increase productivity decreasesby 0.243 in relation to effects in private life all others remaining fixed.
In case we run the last model with the productivity against satisfaction statisticallysignificant variables then the coefficient job satisfaction via education b_1 ¼ 0.617,which implies that the relative risk of this particular variable is eb_1 ¼ 1.853 and thecorresponding percentage change is e b̂1 2 1 ¼ 0.853. This means that in relation to jobsatisfaction via education the odds of persons’ ability to increase productivity increaseby almost 85 per cent ceteris paribus. In the case of job satisfaction via rightnessb_2 ¼ 1.130, which implies that the relative risk of this particular variable iseb_2 ¼ 3.095 and the corresponding percentage change is e b̂2 2 1 ¼ 2.095. Thismeans that in relation to job satisfaction via education the odds of persons’ ability toincrease productivity increases by almost 210 per cent all others remaining fixed.Finally, the odds of persons’ ability to increase productivity increases by 375 per centand decreases by 0.28 per cent in relation to job satisfaction via qualification andorganizations respectively.
The Nagelkerke R square is a measure of predictability of the proposed models(similar to R 2 in a regression). To assess the model fit we compare the log likelihoodstatistic (22 log L̂) for the fitted model with the explanatory variables with this valuethat corresponds to the reduced model (the one only with intercept). The likelihoodratio statistic is quite high in all cases rejecting H0 and concluding that at least one ofthe b coefficients is different from zero.
Finally, the Hosmer and Lemeshow values equal to 4.48, 5.73, 9.85 and 0.92 (withsignificance equal to 0.812, 0.677, 0.276 and 0.969) for the four model formulationsrespectively. The non-significant X2 value indicates a good model fit in thecorrespondence of the actual and predicted values of the dependent variable.
Conclusions and implicationsIn this paper we used factor analysis in order to identify the responsible factors for thecorrelation among a large number of variables and their influence on productivity. Ourresults showed us that productivity is seriously affected by the two qualitative factors,stress and satisfaction. As expected, in the former, increased stress leads to reducedproductivity and in the latter, increased satisfaction leads to increased productivity.
Following this, logistic regression presented us with a lot of useful elementsconcerning the function of stress, satisfaction and supportive elements on productivity.Initially it showed us the effect of financial and social elements such as the importanceof experience and previous employment on productivity, but also the importance of theknowledge that an employee will continue to work for the same organisation, since theincrease of productivity in employees of the same organisation was considerably high.Thus, the trust in older members of staff is a point that can offer a considerableadvantage to the organisation and also a feeling of safety to the employee.
IJPPM59,5
426
Another element that arose was the everyday ordeal concerning getting to and fromwork with this element having a negative effect on employees’ productivity. A problemof decreased productivity also arose in the case of the public sector, which may be true,but at the same time this is connected with lack of motivation, meritocracy, satisfactionetc.
Then the influence of stress on productivity was accentuated, focused on threeelements. First, when work starts to intersect with the workers’ personal life, this has anegative effect on productivity. Second, work load is not connected to the lack ofquality in everyday work, thus quality work is more related to conscientiousness andpersonal satisfaction than work load. And third, energetic and active individuals do notaffect productivity negatively, but positively, and this is why we mentioned the case ofcreative and “useful” stress.
The satisfaction factor greatly affects productivity according to our empiricalfindings. It is important for individuals to work on what they wanted and chose in theirlives, and this is why a large increase in productivity is evident from this element. It ismore important, however, to have a balance between employees’ qualifications andtheir contribution to the organization and the benefits (of all kinds) offered by theorganisation to the employees.
Relying on our sample, we could mention some interesting points. The age andfamily status of the employees is a particularly important factor relating tosatisfaction, because as age increases, the satisfaction from work is reduced, while theyounger the age, the higher the ambition.
In the same way, those who do not have children or are not married find greaterpleasure in work with respect to their free time. With reference to financial situationand education level, we found that workers with high incomes and those with highereducation are more ambitious than other categories. Work experience and in particularyears of work for the same organisation are a stress-reducing factor, since mutual trustbetween organisation and employee contributes to this respect.
With respect to stress and satisfaction, we saw that a large percentage of workersshows stress, but also feels satisfaction from the same organisation. It is noteworthythat the satisfaction ratio is smaller around systems of remuneration and benefits, andso injustices in the remuneration-benefit systems of an organisation may causeconsiderable problems.
Necessary stepsRelying on our empirical results a number of steps are necessary. The following inparticular.
Job description, rotation and redesignA clear job description is needed in order to avoid phenomena of vagueness in roles,fields of action, or role conflicts. The rotation of employees, even horizontally, so thatthey do not reach points where their work is monotonous and boring, is veryimportant. Similarly, change of work areas, if the initial design was incorrect or if theintroduction of changes and re-classifications leads to a recasting of the work area maybe crucial (e.g. the company used to have two employees in the accounts department,but due to development the same space must now accommodate four people).
The effect ofstress and
satisfaction
427
Job environmentThe creation of an environment of understanding and acceptance of such problems bythe company is essential, so that the employee knows that it cares for him and that he isan integral part of the organisation. The existence of recognition and reward for eachwork achievement contributes towards the keeping up of the employees’ morale; in thisway, employees adopt a positive mood towards their role in the organisation and theorganisation shows its members that it does not regard them only as performers, thuscreating a better working climate and reducing the feeling of insecurity and stress.Positive working conditions are considered necessary and non-negotiable factors. Eachworking area as well as the broader environment of the organisation create moods andcontribute towards behaviours and lead to stances.
Cross-functional training and developmentConstant training of the employees is vital not only on matters concerning their workbut also on more general matters concerning the functioning and activities of anorganisation (e.g. seminars on group work, time management, stress management etc.).Similarly greater independent action is required so that the employees can bring outand channel their potential and with the help of management to achieve greatercreativity and innovation.
Job security – safetyWork security and the feeling that employees are not in danger (of remaining unpaid,being fired, being demoted) are important factors. The same applies for control andsupervision, provided, however, it is based on contribution towards better and more justwork attainment and greater group effectiveness, and not on fear of reproach or penalty.
MotivationCreation of a motivator framework which will be renewed and adjusted according toneeds leading to better and more substantial operation of the team may befundamental. This may result also in an increase of team spirit and function as well asto cooperation of management and workers, based on a mutual profitable developmentof the people and the organisation.
Limitations and future extensionsRelying on our findings we have to admit that generalizations to other populationsmust be done carefully. Additional research should focus on the ways to increaseproductivity in public sectors where hearing of an action is difficult to use and hard tomove. The effect of stress and satisfaction on productivity in specific sectors orgeographical areas or professions should also be explored. Finally, more qualitativefactors affecting productivity may be explored like emotional intelligence.
Notes
1. First we define the distributional properties of the dependent variable, (for more details onthe properties and applications of logistic regression see Halkos, 2007; Kleinbaum, 1994;Hosmer and Lemeshow, 1989; Kleinbaum et al., 1999; Hair et al., 1998; Sharma, 1996).
2. We have tried to mix variables of the proposed models but we could not finish with ameaningful model.
IJPPM59,5
428
References
Alderfer, C.P. (1972), Existence, Relatedness, and Growth, Free Press, New York, NY.
Aldwin, C.M. (1994), Stress, Coping and Development. An Integrative Perspective, The GuilfordPress, New York, NY.
Bergus, G.R., Randall, C.S., Winniford, M.D., Mueller, C.W. and Johnson, S.R. (2001), “Jobsatisfaction and workplace characteristics of primary and specialty care physicians at abimodal medical school”, Academic Medicine, Vol. 76 No. 11, pp. 1148-52.
Bogler, R. (2002), “Two profiles of schoolteachers: a discriminant analysis of job satisfaction”,Teaching and Teacher Education, Vol. 18 No. 6, pp. 665-73.
Brewer, E.W. and Clippard, L.F. (2002), “Burnout and job satisfaction among student supportservices personnel”, Human Resource Development Quarterly, Vol. 13 No. 2, pp. 169-86.
Brewer, E.W. and McMahan-Landers, J. (2003a), “Job satisfaction among industrial and technicalteacher educators”, Journal of Industrial Teacher Education, Vol. 40 No. 2, pp. 65-85.
Brewer, E.W. and McMahan-Landers, J. (2003b), “Job stress and burnout among industrial andtechnical teacher educators”, The Journal of Vocational Education Research, Vol. 28 No. 2,pp. 125-40.
Brief, A.P. and Weiss, H.M. (2002), “Organizational behaviour: affect in the workplace”, AnnualReview of Psychology, Vol. 53, pp. 279-307.
Carless, S.A. (2004), “Does psychological empowerment mediate the relationship betweenpsychological climate and job satisfaction?”, Journal of Business and Psychology, Vol. 18,pp. 405-25.
Clugston, M. (2000), “The mediating effects of multidimensional commitment on job satisfactionand intent to leave”, Journal of Organizational Behavior, Vol. 21 No. 4, pp. 477-86.
Dunteman, G.H. (1989), Principal Components Analysis, Sage, London.
Ferrie, J., Marmot, M., Griffiths, J. and Ziglio, E. (Eds) (1999), Labour Market Changes and JobInsecurity: A Challenge for Social Welfare and Health Promotion, WHO RegionalPublications, Copenhagen.
Freud, S. (1978), I and the Defence Mechanisms, Kastaniotis Publications, Athens, (translationParadellis Th.).
Gay, L.R. and Airasian, P. (2003), Educational Research: Competencies for Analysis andApplication, 7th ed., Prentice-Hall, Upper Saddle River, NJ.
Greene, W. (2000), Econometric Analysis, Prentice-Hall, Englewood Cliffs, NJ.
Gurmu, S., Rilstone, P. and Stern, S. (1999), “Semiparametric estimation of count regressionmodels”, Journal of Econometrics, Vol. 88 No. 1, pp. 123-50.
Hair, J.F., Anderson, R.E., Tatham, R.L. and Black, W.C. (1998), Multivariate Data Analysis,Prentice-Hall, Englewood Cliffs, NJ.
Halkos, G.E. (2007), Econometrics: Theory and Practice, Giourdas Publications, Athens.
Halpern, R. (1999), “After school programs for low-income children: promise and challenges”,Future of Children, Vol. 9 No. 2, pp. 81-95.
Herzberg, F. (1966), Work and the Nature of Man, Work Publishing Co, Cleveland, OH.
Heslop, P., Smith, G.D., Metcalfe, C. and Macleod, J. (2002), “Change in job satisfaction, and itsassociation with self-reported stress, cardiovascular risk factors and mortality”, SocialScience & Medicine, Vol. 54 No. 10, pp. 1589-99.
Hosmer, D.W. and Lemeshow, S. (1989), Applied Logistic Regression, John Wiley and Sons,New York, NY.
The effect ofstress and
satisfaction
429
Johnson, R.A. and Wichern, D.W. (1998), Applied Multivariate Statistical Analysis, Prentice-Hall,Englewood Cliffs, NJ.
Karasek, R.A. (1979), “Job demands, job decision latitude, and mental strain: implications for jobredesign”, Administrative Science Quarterly, Vol. 24 No. 2, pp. 285-308.
Kleinbaum, D.G. (1994), Logistic Regression: A Self Learning Text, Springer-Verlag, New York,NY.
Kleinbaum, D.G., Kupper, L.L., Muller, K.E. and Nizam, A. (1999), Applied Regression Analysisand Other Multivariate Techniques, Duxbury Press, Belmont, CA.
Labiris, G., Gitona, K., Drosou, V. and Niaks, D. (2008), “A proposed instrument for theassessment of job satisfaction in Greek Mental NHS hospitals”, Journal of Medical Systems,Vol. 32 No. 4, pp. 333-41.
Leeflang, P., Wittink, D., Wedel, M. and Naert, P. (2000), BuildingModels for Marketing Decisions,Kluwer Academic Publishers, Boston, MA.
Locke, E.A. (1976), The Nature and Causes of Job Satisfaction, Rand McNally, Chicago, IL.
Maslow, A.H. (1954), Motivation and Personality, Harper, New York, NY.
Paoli, P. (1997), Working Conditions in European Union, European Foundation for Living andWorking Conditions, Dublin.
Petty, G.C., Brewer, W. and Brown, B. (2005), “Job satisfaction among employees of a youthdevelopment organization”, Child and Youth Care Forum, Vol. 34 No. 1, pp. 57-73.
Porter, L.W. and Lawler, E.E. (1986), Managerial Attitudes and Performance, Dorsey Press,Homewood, IL.
Selye, H. (1964), The Stress of Life, Mc Graw-Hill Book Co., New York, NY.
Sharma, S. (1996), Applied Multivariate Techniques, John Wiley and Sons, New York, NY.
Siegrist, J. (1996), “Adverse health effects of high-effort low-reward conditions”, Journal ofOccupational Health Psychology, Vol. 1 No. 1, pp. 27-41.
Warr, P. (1990), “Decision latitude, job demands, and employee well-being. Work and stress”,Community of Scholars, Vol. 4, pp. 285-94.
Further reading
Catalano, R. (1991), “The health effects of economic insecurity”, American Journal of PublicHealth, Vol. 81 No. 9, pp. 1148-52.
Dooley, D., Fielding, J. and Levi, L. (1996), “Health and unemployment”, Annual Review of PublicHealth, Vol. 17, pp. 449-65.
Elkedes, T. and Seifert, W. (1993), “Unemployment and health impairments longitudinalanalyses for the Federal Republic of Germany”, European Journal of Public Health, Vol. 3No. 1, pp. 28-37.
Fenwick, R. and Tausig, M. (1994), “The macroeconomic context of job stress”, Journal of Healthand Social Behavior, Vol. 35 No. 3, pp. 266-82.
Graetz, B. (1993), “Health consequences of employment and unemployment: longitudinalevidence for young men and women”, Social Science andMedicine, Vol. 36 No. 6, pp. 715-24.
Heaney, C., Israel, B. and House, J. (1994), “Chronic job insecurity among automobile workers:effects on job satisfaction and health”, Social Science and Medicine, Vol. 38 No. 10,pp. 1431-7.
Herzberg, F. (1968), Work and the Nature of Man, World, New York, NY.
IJPPM59,5
430
Leino-Arjas, P., Liira, J., Mutanen, P., Malvivaara, A. and Matikainen, E. (1999), “Predictors andconsequences of unemployment among construction workers: prospective cohort study”,British Medical Journal, Vol. 319 No. 7210, pp. 600-5.
Lewis, M. and Ramsay, D.S. (1997), “Stress reactivity and self recognition”, Journal of ChildDevelopment, Vol. 68 No. 4, pp. 621-9.
Tennant, C. (2001), “Work-related stress and depressive disorders”, Journal of PsychosomaticResearch, Vol. 51 No. 5, pp. 697-704.
Turner, J. (1995), “Economic context and the health effects of unemployment”, Journal of Healthand Social Behavior, Vol. 36 No. 3, pp. 213-29.
Viinamaki, H., Koskela, K. and Niskanen, L. (1993), “The impact of unemployment onphychosomatic syptoms and mental well-being”, The International Journal of SocialPhychiatry, Vol. 39 No. 4, pp. 266-73.
About the authorsGeorge Halkos is an economist who graduated from the Department of Economics at theUniversity of Piraeus (Greece). He holds a Master of Sciences (M.Sc) in Project Analysis, Financeand Investment and a Doctorate of Philosophy (D.Phil) in cost-benefit analysis with the use ofmathematical models and game theory and with application to environmental economics fromthe Economics Department of the University of York (England). He has worked as team leaderand research fellow in various research and academic institutions, such as the University ofPiraeus, the Athens University of Economics (former ASOEE), the University of the Aegean, theStockholm Environment Institute, the Panteion University, the ATEI, for research projectsfinanced by the European Union (PHARE programs), the Hellenic Republic Ministry ofDevelopment, the National Statistical Service of Greece, the EUROSTAT, the Ministry ofEmployment. He has participated and presented articles in international conferences organisedby the ENI, the Centre of European Economics of London, the Royal Economic Society. He acts asa referee in scientific journals such as the Energy Economics, Environmental and ResourceEconomics, Empirica, Environment and Development Economics, Ecological Economics, Journalof Industrial Ecology, Journal of Environmental Management, Greener ManagementInternational (The Journal of Corporate Environmental Strategy and Practice), Research inInternational Business and Finance, European Journal of Operational Research, AppliedMathematics and Computation. He is the Director of Postgraduate Studies and Deputy Head inthe Department of Economics at the University of Thessaly entitled “Applied Economics”. He isalso the Director of the Laboratory of Operations Research. His research interests are in the fieldsof applied statistics and econometrics, simulations of economic modelling, environmentaleconomics, applied micro-economic with emphasis in welfare economics, air pollution, gametheory, mathematical models (non-linear programming). George Halkos is the correspondingauthor and can be contacted at: [email protected]
Dimitrios Bousinakis graduated from the University of Piraeus and received a degree inBusiness Administration. He holds an MBA and graduated from the National TechnicalUniversity of Athens (with the coordination of University of Piraeus and University of Athens,with a major in Production Management) and a PhD from the University of Thessaly,Department of Ecomonic Studies (with a major in Managerial Psychology). He is a financialanalyst in MS Jacovides SA (a medical device company).
The effect ofstress and
satisfaction
431
To purchase reprints of this article please e-mail: [email protected] visit our web site for further details: www.emeraldinsight.com/reprints
